SCANNING ELECTRON MICROSCOPY OF PLYWOOD GLUE LINES1 Zoltan Koran Faculty of Forestry, University of Toronto, Toronto 5 and Ramesh C. Vasishth Reichhold Chen~icals( Canada) Ltd., Toronto 5 (Received 1 October 1971)

ABSTRACT Douglas-fir plywoocl panels made under specific conditia~ns to represent dried-,z~ut. i~ndercured,high-moisture, and rough-veneer glue bonds were tested in shear, and the failure surfaces were ev:~luatedby light and scanning electron microscopy. Results showed low shear strength and low percentages of wood failure in all abovc panels. Wood failure occurred both by intrawall failure and transwall failure parallel with the gi.ain directiol~ in wood. Scanning electron micro!icopy made it possil>le to distinguish between glue-coated areas versus uncoated wood fracture and clearly showetl glue distribution and th~.physical form of glue on the failure surface. Dried-out bonds are c,haracterized by the lack of a contin~~onsglue line, little or no transfer of glue onto the face veneer, and mostly by adhesive failure. The most important features of high-moisture joints are the formation of numerous steam bubbles in the glut- line and the lack of a continuous glue bond. These are due to the excessive sqneeze.out of glue in the press and the overpenetration of glue into the wood. Both rough-ve11ee1 and low-pressure joints provide discontinnous glue lines because of the lack of continuoil\ contacts between the adjacent veneers. All results show that examination of shear specimens by SEM is an effective metl~otl for determining the cause of glue-line failure.

INTRODUCTIO Y ferently from wood, either on its own or ~h~ study of glue lines has been of prime through the atldition of fluorochrornt:s. interest to the plywood ma~iufacturerat the The present study employs scanning manufacturing plants. The quality of glue "l"ct'0" microscopy for glue lin': c'valu- lint: is evaluated mostly l,y the standard ation. This instrunlent has been selected shear test, and the cause of failure is de- because it is able to distinguish inost ac- terminc, mostly on the bases of visual curately between the areas covercd with observations of the exposc:d fracture sur- glue and those that are free of adhesives. faces. At the research lev(:], the glue line MATERIALS AND METHODS is routinely studied by light microscopic This investigation utilized Douglns-fir techniques, including standard stereo and plywood panels formed with phenol form- transniitted light, and more recently by aldehyde for the study of the quality of fluorescence microscopy in sectional and glue bond. several panels were surface views (Quirk 1968; C6t6 and esperimcntal]y with the following vari- Vasishth 1970). The latter is based on the ables: ( 1) high moisture contellt ( 12 % ), property that the adhesivct fluoresces dif- (2) dried-out glue line before hot press- ing, (3) rough-veneer surface, (4) short lThe support of this resea~ch by a National pressing time, and (5) normal conditions. Ind~lstrialResearch Aid Grant is greatly appreci- The of glue bond was evaluated ated. The plywood panels were prepared and by standard shear test and per- tested by W. C. Ainslie and W'. Srnith at Reich- f hold Chemicals (Canada) Ltd. Their contri- centage of wood failure. In each case the butions, too, are greatly appreciated. onc-square-inch shear failure surface was 202 SCANNING ELECTRO~~~ICROSCOPY OF ~'LYWOI~I)GLUE LINES 203 studied by stereo-light microscopy, and the Good bond percentage of wood failure was determined This bond is formed as a result of by area measurements. optimum conlditions in wood, glue, and in Representative areas (1 cm" were also the manufacturing variables. Good bond is selected for scanning electron micro- characterized by high breaking stress and scopic observations. These specimens were by a high percentage of wood failure mounted onto standard aluminum stubs (93% ) obtained in the shear test. hlficro- with double-coated tape and were then scopic examination shows continuous glue coated with a 200-A-thick layer of gold- lines between the various veneer layers, palladium alloy in a high vacuum evapo- without the presence of air bubbles, and rator. The edges of the specimens were optimum glue penetration into the tracheid painted with silver to provide contact with lumens immediately adjacent to the glue the aluminum stub and the examina t' ion line. Here the glue remains in close contact was carried out in the Cambridge Stereo- with the lurnen lining without any sepa- scan I1 electron microscope. This technique ration between the cell wall and the glue. provided excellent three-dimensional views As expected from the geometry of the of the overall surface morphology of ply- test specimen, the plane of failure coin- wood fracture. cides with the tangential plane in wood. More specifiually, the zone of failure falls RESULTS AND DIS(XSSI0NS either in the springwood zone, or alter- Failure can occur in principally three natively at the growth ring boundary. At diffcrcnt sites: (1)in wood, (2) within the the cellular level the fracture path may go glue line (cohesive failure), and (3) at across the tracheids, or it may remain the interface of wood and glue (adhesive within the tracheid wall. failure). When failure occurs entirely Figure 1 slhows an example in which the within wood, thus exposiilg wood on both tracheids on the veneer surface arc. split surfaces, this is designated as "100% wood into two hal-rres parallel to their long axes, failure."' This occurs mostly when there is thus exposing the fractured radial walls a good glue bond, but sometimes when the (R) and the inner surfaces of tracheids veneer surface is weakened by frequent from the lurnen side, that is the S3 layer and intensive lathe checbing. of the secondary wall with the spiral Conversely, when failul-c. occurs entirely thickening (L,). In the fractured radial within the glue line, without exposing any walls (R), all layers of the fiber wall are wood, this is designated as "cohesive glue exposed. This type of wood failure is failure." This type of failure is due to defined as "transwall failure parallel with inferior glue quality, since under normal the tracheid axis." conditions a good glue bond is stronger Figure 2 shows a representative example than wood. where failure occurs across several tra- When failure occurs at the interface of cheicls, as a result exposing their end walls the glue line and wood, thus exposing and lumens. 'This type of failure is defined 100 % glue on one surface and 100 % wood as "transwall failure perpendicular to the on the opposite surface, this is designated tracheid axis" ( Koran 1968 ) . as "adhesive failure." This occurs when Figure 1 shows several examples in thc adhesive strength between glue and which the site of failure falls within the wood is weaker than either the cohesive double tracheid wall (I), thus exposing strength of glue or the strength of wood. its internal structure. This type of failure In most instances failure in the glue line has been defined as "intrawall failure" region involves the combination of the (Koran 19613). In general, the areas ex- three types of failure: wood, glue, and posed by transwall failure parallel with the interface. These will be described in the tracheid axis and by intrawall failure are various types of glue joints. about equal. Transwall failure perpendic- 204 ZOLTAN KORAS AND RAMESFI C. TTAS19HTII SCANNISG ELECTRON MICROSCOPY OF PLYWOOI) GLUE LINES 205 ular to the tracheid axis is the least com- veneer. In Fig. 4, for example, the parallel mon type of failure. microfibrillad structure of the S1 layer is accurately r &produced,while in Fig. 6 a Dried-out glut line cast of the 1ibcr lumen is produced with This type of glue line is formed when the inner subface of the lumen lining ac- much of the water has evaporated from curately reQlicated. These micrographs the glue film before the glue film contacts suggest that lin these areas fairly close con- the unspread surface. As a result, the glue tacts have been established between the film does not flow under heat and pressure wood and glue, but probably specific ad- and adequately "wet" the unspread surface. hesion was ]lot produced because of the Thus, a substandard glue bond is formed. inability of l:he prehardened glue to wet Studies made on four independent the fiber surface. "dried-out" specimens revealed ( 1) low Unbonded area represents the glue sur- shear stress values, (2) low percentage of face without any imprint of the face veneer wood failure (15% ), (3) little or no glue in it. In the e regions, no contact has heen transfer, (4) little or no glue penetration established 1between the face veneer and into the face veneer. Bond failure con- glue spread core. Therefore, in the shear sisted mainly of (a) wood failure, (b) tests no failure occurs in these regions. adhesive failure, and (c) unbonded areas. Scanning elektron micrographs of the, glue A representative example of wood failure spread core veneer (Figs. 3 and 4 at G) is shown in Fig. 3. Fragnicmts of tracheids reveal that time areas possess undisturbed (W) torn out of the face veneer remain and smooth glue-covered veneer surfaces, attached to the glue on the core surface. except for tde variations in overall surface This means that in these areas, either the topography. Under incident light illumi- wood was weakened, or thc. glue bond was nation, thesi? glue-coated areas appear stronger than the wood itself. Figure 3 also highly shiny and somewhat glazed. shows that wood failure occurred in the The presejierl of unbonded areas in the forms of intrawall failuicb and transwall glue line ha/s been confirmed by Fig. 5, failure parallel with the tracheid axis. which was tqken from an area where there Adhesive failure can be identified by the was a knothcjltl in the face veneer. As a re- sult this was an open glue line on the core inlpression the face veneer leaves in the without any contact with the face vcweer. glue on the core surface. Representative As expected,,the surface morphology of the examples are seen in Figs. 4 and 6. These glue line in he knothole area (Fig. 5) was imprints are fairly accurate, replicas formed identical to t,hose4 in the unbonded areas of of the tracheid surfaces exposed on the face dry-out glue joint (Figs. 3 and 4).

FIG. 1. Scanning electron micrograph (SEM) of Douglas fir plywood failure in the tangential plane showing transwall failure parallel with the tracheid axis, traliheids with spirals (L),and intra- wall failure (I)indicated by tracheids without spirals. All scales are in micrometers (pm). FIG. 2. Transwall failure perpendicular to the tracheid axis elcposing the cross walls and lumens of tracheids (A). FIG. 3. Dried-out glue joint showing wood failure (W) and ad/iesive failure consisting mo\tly of unbonded areas ( G ) . FIG. 4. Dried-out glue line showing adhesive failure with illprints of the tracheid wall (S,), 1111boncled areas (G) and steam vents (V). FIG.5. Dried-out glue lint: consisting entirely of unbonded gl+ on the core veneer just below a knothole in the face veneer. Kote that the glue exposes smooth aqld undisturbed surface topography. FIG. 6. Dried-out glue line, in which the spiral thickenings of I>onglas-fir tracheids are accurately replicated. 206 ZOLTPLN KORAN AND RIUIESH C. VASIllEITH SCANKING ICLECTRON M~CROSCOPY OF PLYWOOD GLUE LINES 207

Starved glue ioint complete, but since most of the glue pene- This glue line was formed of veneer with trated into the veneer surface, little glue high-moisture content which caused much rcmained to form the final glue boncl. of the glue to penetrate into the veneer Another common feature of the high- surface, thus leaving an insufficient amount moisture glue bond is the formation of of glue to form a continuous glue line. steam pockets in the glue line during the Studies made on panels possessing starved hot press operation at 340 F. Although gluc joint revealed: (1) low shear stress much of the steam escapes during the values, (2) low percentage of wood failure pressing operation, part of the steam be- (12% ), (3) numerous blisters, (4) the comes captured within the glue line in absence of a continuous glue line, and (5) steam pockets, such as the one observcad in the presence of deep glue penetration into Fig. 8. These pockets vary in sizc and the wood, but not enough glue to fill com- shape (Fig. 8), and the larger ones nlay pletely the openings on the veneer surface. develop into blisters upon the release of The deep glue penetration was especially pressure. Other forms of steam pockets are apparent in thc wood rays where glue observed in Fig. 9. In such cases, the entire penetration was several cc-11s deep. This is glue line seems to possess a uniform foam- illustrated in the scannin~electron micro- like texture, indicating the presence of an graph of Fig. 7 where the polelike struc- excessive amount of steam throughout the turcs in the center of the picture are solid glue bond during pressing. gluc bodies pulled out of the ray cells from the opposite surface during the shear test. Untlercured glue line In Fig. 7 the tall glue pole is 45 pm in In undercured plywood panels. the height, which means a minimum of 45 pm formation of glue bond is normal up to penetration into the opposite surface. In the pressing operation. However, in the general, glue penetration was deeper into hot press, the pressure is released I)efore the core surface than into the face veneer. the complete polymerization of the glue is Microscopic examination of the exposed achieved, that is, before maximum strength fracture surfaces revealt~l uniform glue is developed in the glue bond. Therefore, coverage on both the face and core veneer in undercurtxl plywood panels, tht, glue surfaces. However, these :jurfaces exhibited bond fails in many areas upon the release a highly washed-out appearance, with little of pressure. 'This is because the partially glue remaining on the sur.face. This means cured glue bond is not able to withstand that glue transfer from thc: spread core onto the localized tensile forces upon the spring- the unspread face venecr was generally back of the veneer.

FK:. 7. Glrlr liric formed on veneer \\,it11 high-moisture content. The depth of glue penvtration is indicated by the height of gh~ecasts (G) pullecl out of some ray cells in the shear test. All scales are in micrometers ( pm ) . FIG. 8. Glut line formed c~fhigh-moistnrc vcmecr. Cohesive failnre (C) in glne is clearly apparent l,etn,een the steam pockets, \vluch weaken the glue line. FIG. 9. C:l~ic line for~netl of veneer with high-moisture content, showing foamlike structllre ant1 collesi\-e failure between the :;team pockets. FIG. 10. Wood failure (U') and glue failure (G)in undercured glue line. Note also the ac- cum~llationof glue in the lathe check (L). FIG. 11. Rough veneer glue line showing the accumulation of glue in the lathe check (I,), glue failnre (G) ancl wood fai111rc. ( W). FIG. 12. Transverse co~llpression folds (C) in the walls of tracheids immediately adjacent to the glue line which causes their formation as a result of existing compressive stresses. 208 Z~LTANKORAN AND RAI~IESII C. VASISHTH

In this study the undercured plywood in Figs. 4 and 6. These probably origiliatecl panels posscssed low shear strength and a from the intermediate pressure areas in the low percentagc (50%,) of wood failure. plywood, where pressure was riot high SEhl observations revealed an equally uni- enough to establish a complete bonding form glue coverage and penetration on between wood and glue. both veneer surfaces and the presence of a In general, plywood formed of rough continuous glue line. Figure 10 is a low- veneer possessed low shear strength and magnification view of an undercured glue low percentagc of wood failurc (15:Z ). bond exposing surfaccs of both wood (W) and gluc ( G ) failurc. Closer examination CONCLUSIONS of the glue failure reveals both cohesive Plywood p:inels formed of veneer with and adhesive failures. high-moisture contents, rough surface Figure 12 shows an area containing gluc topography, dried-out glue on thc core failure (G) and wood failurc (W). In the surface, and undercured glue linc, are latter, transvcrse compression folds (C) characterized by low shear strength and can be observed in the tracheid walls im- low percentage of wood failure. ~nediatelyabove the glue line. These folds wcre introduced during hot pressing and Good bonds possess a continuous glue arc now maintained by the gluc line just line that is stronger than the wood itself. In shear test this type of bond produces below the wood surface. This proves that ncarly 100% wood failure, which occurs localized stresses are set up during the preferentially in springwood or at the pressing operations and tl~esc,stress zones growth ring boundary. Wood failure occurs are maintained in the pl~woodpanels by in the forms of intrawall failure, trans- the resisting effect of the qluc line. wall failure - parallel with the grain direc- tion, and transwall failurc ~erpendicularto Rough veneer joint the tracheid axis. Failure types of the first In many respects the glue line of ply- two kinds arc, of about equal occ:urrence, wood panels formed of rough veneer is while the last is rather rare. similar to thc dried-out gluc line. When Dried-out bonds are characterizt d by the glue is spread onto the rough core veneer lack of glue transfer, glue flow, wetting surface, it tends to flow into the depres- sions, cracks, lathe checks (Fig. 11 at L), and penetrati~on, and by the lack of a accumulate in these area\ arid leavc rel- continuous gluc line. The shear fracture atively little glue in the high areas. In surface of dried-out bond displays high the hot press, direct contact is made only percentages of adhesive failure and un- in the high areas wherc some bond is bonded arcas and a low percentagt3 ( 15% ) formed, while in the depressions where of wood f a1'1 ulre. most of the gluc accun~ulates,no contact is Glue joints fonlled of veneer \\,it11 high made. Here, the glue simply dries onto the inoisture content reveal good glue transfer core veneer surfacc ancl cures without from core onto face, overpenet~ation of fonning any bond. This type of panel gluc into wood, excessive squeeze-out of possesses starved localizcd joints in the glue in the hot press, the forination of high areas and dried-out. unbonded glue numerous air bubbles in a rather dis- line in the low areas. continuous, starved glue line, and a low In many cases the glue in the unbonded percentage of wood failure ( 12% ). areas contained large bubbles, similar to Shear failure surfaces of rough veneer those in Figs. 8 and 9. In other areas of the joints arc characterized by adhesi\.e f ;ii'1 ure fracture surface, impressions of the face with imprints in the glue, cohesive fa'ii 'I urc veneer were formed in the glue surface of in glue, unbonded glue areas, and by a the core veneer, such as the type observed low percentage of wood failure ( 157; ). SCANNING E:LECTHOX AfICROSCOl'Y OF PLYWOOD GLUE LINES 209

REFERENCES gential tracheid surfaces of black spruce prodllced by tensile failure at various teinper- Ci,Ti,W, *., Asu R. C. VASISHTH. 1970. Light and electron microscopic: studies of wood/ atures. Sven. Paperstidn. 71 ( 17): 567-576. wood coating systems. 10th Fatipec Congress. QUIRK, J. T. 1968. Fluorescence microscopy for Verlag Chemie 6mbH, WeinheimIBergstr. p. detecting adhesives on fracture surfaces. 57-65. USDA Forest Serv. Forest Prod. Lal). Res. KORAN, Z. 1968. Electroil niicroscopy of tan- Note FPL.-0191.